David L. Streiner, PhD, of McMaster University, Hamilton, Ont., and the University of Toronto, joins Blood & Cancer host David Henry, MD, of Pennsylvania Hospital, Philadelphia, to explain what P values actually measure and how they both help and hinder the interpretation of clinical research findings.
Plus, in Clinical Correlation, Ilana Yurkiewicz, MD, of Stanford (Calif.) University, explores how quickly cancer can turn into bankruptcy.
Show notes
- In statistics, P value is null hypothesis significance testing.
- The P value assesses the following: If the null hypothesis (i.e., there is no difference) is true, what is the probability that we could get data that is extreme?
- What are researchers doing when they test this way? Given the null hypothesis (i.e., we are assuming data is from chance alone), what is the probability that the data are actually true?
- What do researchers actually want to be able to do? Given the data, what is the probability of the null hypothesis (i.e., random chance alone is responsible for the difference)?
- The P value is affected by sample size; a smaller sample is more easily influenced by variable data and can result in outcomes that are not statistically significant. Large sample sizes are affected less by variables.
- It is important to differentiate what is statistically significant from what is clinically significant.
- Remember, P less than .05 is an arbitrary number. Do not let a P value deter use of a therapy that may show clinical benefit.
Show notes by Ronak Mistry, DO, resident in the department of internal medicine, University of Pennsylvania, Philadelphia.
For more MDedge Podcasts, go to mdedge.com/podcasts
Email the show: podcasts@mdedge.com
Interact with us on Twitter: @MDedgehemonc
David Henry on Twitter: @davidhenrymd
Ilana Yurkiewicz on Twitter: @ilanayurkiewicz
Podcast Participants
David Henry, MD